Toward inclusive healthy ageing: a systematic review of equity in healthy ageing interventions Authors: Chava Kurtz, Odelia Ben Harush, Anna Zisberg, Paola Buedo, Stuart McLennan, Efrat Shadmi Published: 2026 ABSTRACT BackgroundHealth inequities arising from social determinants such as income, education, gender, and socioeconomic status profoundly affect disease burden and ageing outcomes. Despite global efforts to promote equitable healthy ageing, substantial gaps remain. This systematic review examined how equity-related factors, defined by the PROGRESS framework: Place of residence, Race/ethnicity, Occupation, Gender/sex, Religion, Education, Socioeconomic status, and Social capital, are incorporated into evaluations of healthy ageing interventions and how outcomes vary across these dimensions.MethodsPubMed, Embase, and Scopus were searched from database inception through August 12, 2024, for quantitative evaluations of healthy ageing interventions.ResultsOf 5,892 screened studies, 66 met inclusion criteria, most published after 2018 and involving older adults (mean age 71.4 years). The most frequently reported PROGRESS factors were gender/sex (N = 66, 100%), education (N = 49, 74%), and race/ethnicity (N = 38, 58%). Twenty-five studies (38%) reported PROGRESS-related data without analysing outcomes by PROGRESS factors, whereas 41 studies (62%) analysed intervention outcomes by at least one PROGRESS dimension. Across 254 extracted outcome results, 98 (38.6%) suggested decreased inequity, 146 (57.5%) were non-significant, and 10 (3.9%) suggested increased inequity. Physical activity interventions were the most frequently studied and showed the largest number of favourable findings, whereas social connectedness interventions showed the highest proportion of unfavourable outcomes. Intersectional analyses were rare, and only six studies examined more than one PROGRESS dimension jointly.ConclusionsEquity considerations are increasingly present but remain inconsistently integrated into healthy ageing intervention research. Greater attention to underrepresented PROGRESS dimensions, improving standardised equity reporting, and expanding intersectional analyses is needed to ensure interventions contribute to reducing, rather than reinforcing, inequities.RegistrationThe protocol was registered with PROSPERO (CRD42024585709) before the final search.Supplementary informationThe online version contains supplementary material available at 10.1186/s12939-026-02905-z. Background Population ageing is a defining demographic trend worldwide. The number of people aged 60 and older is projected to increase from 1.1 billion in 2023 to 1.4 billion by 2030 [1]. This demographic shift is accompanied by rising multimorbidity, commonly defined as a coexistence of two or more chronic conditions in the same individual [2], as well as increasing physical, cognitive and functional dependency [3]. Yet current healthcare systems remain largely organized in professional and organizational silos and are poorly aligned with these complex realities. Fragmented health and social care contribute to poor patient outcomes [4]. These challenges directly hinder the achievement of healthy ageing, defined as “the process of developing and maintaining the functional ability that enables well-being in older age” [5]. Promoting healthy ageing requires creating the environments and opportunities that enable people to be and do what they value throughout their lives [6]. The World Health Organization (WHO) identifies several strategies to promote healthy ageing, including community-based support initiatives, promoting mobility and independency, and strengthening integrated, person-centered health care [7]. In the last decade, interventions to support older adults to age actively have proliferated, particularly those targeting physical activity, self-management, and social functioning [8]. Previous evidence syntheses have examined several aspects of these interventions, including implementation approaches for healthy ageing interventions, as well as community-based interventions designed to sustain independence and improve older adults’ physical health, psychological wellbeing, and social connections [9, 10]. Collectively, these reviews suggest that healthy ageing interventions may have positive effects across multiple domains, although findings are heterogeneous and some intervention models appear more beneficial than others. Contexts in which people live and work shape disease risk and influence healthy ageing [11]. Health inequities, defined as “differences in health status or the distribution of health determinants between population groups” [12], often manifest as barriers to accessing care and as greater risk of poor health outcomes [13, 14]. Addressing ethical, legal, and social implications is essential to advancing equity in healthy ageing, by ensuring that interventions are designed and implemented in ways that are fair, inclusive, and responsive to diverse needs. Tackling inequities driven by the social determinants of health, such as socioeconomic position, gender and place of residence, is essential to ensuring that healthy ageing is inclusive and enables all groups to attain the health outcomes currently experienced by more advantaged populations [11]. Nonetheless, healthy ageing interventions may not be equally implemented among population groups, and may, in some cases, inadvertently widen existing inequities [15, 16]. This concern is particularly salient in the context of intersectionality, where overlapping social determinants of health interact to compound disadvantage [17]. Some equity-focused syntheses have emerged in related ageing fields. For example, a systematic review of mHealth interventions for caregivers of older adults applied a PROGRESS-Plus framework to examine how equity was considered in intervention design, implementation, and evaluation [18]. However, previous reviews have not specifically examined the extent to which studies of healthy ageing interventions in public health and healthcare settings incorporate equity considerations in data collection, reporting, and analysis across PROGRESS dimensions, nor whether these interventions reduce, have no differential effect on, or increase inequities across social groups. Evidence is also limited regarding whether intersectional combinations of disadvantage are examined as they relate to the implementation and outcomes of healthy ageing interventions [19, 20]. To address this gap, this review assesses whether healthy ageing intervention studies collected, reported, and analysed data on eight factors characterizing the social determinants of health, termed - PROGRESS factors: Place of residence, Race/ethnicity/culture, Occupation, Gender/sex, Religion, Education, Socioeconomic status, and Social capital [11]. In addition, we examine whether intervention effects differed across these dimensions and their intersections, and whether interventions decreased inequities, had no differential effects, or increased inequities among populations characterized according to PROGRESS characteristics. Methods Search strategy and selection criteria This systematic review was carried out following the PRISMA-Equity 2012 Extension for systematic reviews with a focus on health equity (PRISMA-E) [21, 22] (Supplementary File 1). The study protocol was registered on the PROSPERO international prospective register of systematic reviews (registration number: CRD42024585709) before the final search was performed. The PROGRESS framework [11] was used to characterise dimensions of inequity and to focus on the core social determinants captured within the framework. To examine intervention effects in relation to inequity, studies that analysed results according to PROGRESS factors were identified, and subgroup-specific findings were extacted where reported. This included both studies comparing outcomes across PROGRESS-defined subgroups and studies conducted in populations defined by one or more PROGRESS characteristics. For studies comparing PROGRESS-defined subgroups, both the overall effect in the total study population and the reported subgroup-specific outcomes were examined. Studies in the second category were those in which the entire study population was characterised by one or more specific PROGRESS factors, for example, interventions delivered to people of low socioeconomic status. To classify interventions, given the wide heterogeneity of studies included in this review, the World Health Organisation’s International Classification of Health Interventions (ICHI) [23, 24] was used to define intervention types. This classification provides a standardised framework describing interventions across the full scope of health systems, including prevention, health promotion, treatment, and rehabilitation activities [24, 25]. This approach allowed consistent coding and comparison across clinical, public health, and policy contexts. According to the ICHI, the included studies were grouped into six intervention targets categories for synthesis: (1) physical activity, (2) lifestyle behaviours, (3) social connectedness, (4) psychological wellbeing and coping, (5) medication use and healthcare engagement, and (6) other specified targets. These interventions were assessed against three outcome categories: (1) patient-reported outcomes, including quality of life, perceived functioning, and self-efficacy; (2) clinical outcomes, including biomarkers, physical performance, and mortality; and (3) healthcare use. This review formed part of a broader study investigating the ethical, legal, and social implications of healthy ageing. The initial search strategy was developed in collaboration with research specialist librarians at the University of Haifa and was subsequently refined by the research team. The search strategy was structured around three concept clusters: healthy ageing, public health/policy interventions, and ethical, legal, and social implications (ELSI)/equity-related concepts. Synonyms and related terms within each cluster were combined using the Boolean operator OR, and the three clusters were combined using AND. The healthy ageing cluster included terms such as “healthy aging”, “healthy ageing”, “successful aging”, “successful ageing”, “aging”, and “ageing”, together with terms including “healthy lifestyle”, “quality of life”, “mental health”, “multimorbidity”, “intrinsic capacity”, “disability”, and “comorbidity”. The public health/policy cluster included terms such as “public health”, “health policy”, “healthcare delivery”, “program*”, “guideline*”, “intervention*”, and “strateg*”, while the ELSI/equity cluster included terms such as “ethic*”, “autonomy”, “justice”, “equity”, “inequality”, “discriminat*”, “ageism”, “legal”, and “social determinants of health”. Controlled vocabulary terms and free-text terms were adapted for each database. We searched PubMed, Embase, and Scopus from database inception through August 12, 2024. We also conducted backward and forward citation searching of the included studies to identify additional relevant papers. The full database-specific search strategies are provided in Supplementary File 2. Eligibility criteria were defined a priori. Peer-reviewed quantitative intervention studies reporting subjectively or objectively measured outcomes of interventions delivered in public health or healthcare settings and aimed at promoting healthy ageing or preventing adverse ageing-related outcomes, including multimorbidity, were included. Studies were eligible if outcomes were reported overall and or stratified by at least one PROGRESS factor. Reviews, conference abstracts, study protocols, and non-human studies were excluded. Search results across all included databases were imported to the Covidence systematic review software (Veritas Health Innovation) and duplicate studies were removed. Studies were screened by title and abstract independently by two members of the research team (any two of the following: OBH, CK, AZ) based on the predetermined inclusion criteria. Studies were screened to assess whether they examined interventions relating to healthy ageing, and those were further assessed to see which examined aspects of equity according to the PROGRESS framework. Full texts of potentially eligible studies were screened independently by OBH and CK. Disagreements between reviewers during the screening process regarding the inclusion/exclusion of studies were resolved by discussion with ES or AZ. Reasons for exclusion of a study were recorded as part of the full-text screening process. Data analysis Data extraction Data extraction was performed independently by two reviewers (any two of the following: OBH, CK, ES), with discrepancies resolved through discussion. For all included studies, bibliographic details, study design, study aim(s), participant characteristics, main intervention characteristics, and study outcomes were extracted. Information was also recorded on whether and how PROGRESS factors were used into analyses of healthy ageing intervention effects. Studies that did not report intervention outcomes according to PROGRESS characteristics were classified into two groups: those that reported PROGRESS characteristics only in baseline descriptions of the study population, and those that included PROGRESS variables solely as covariates in multivariable models, without examining or reporting differential outcomes across subgroups. For studies that examined differential intervention effects according to at least one PROGRESS factor, a more detailed extraction form was used. Data were extracted separately for studies comparing outcomes across PROGRESS-defined subgroups and for studies conducted exclusively in populations defined by one or more PROGRESS characteristics. For studies comparing PROGRESS-defined subgroups, data on both the overall intervention effect in the full study population and the subgroup-specific results were extracted, where reported. When analyses were limited to pre-post assessments within groups, and no between-group comparisons were provided, within-group results were extracted. When between-group comparisons were available, for example through difference-in-difference analyses, only the between-group results were extracted. For studies reporting follow-up outcomes, the follow-up timepoint was treated as the primary endpoints and used in the analyses. Quality assessment The methodological quality of the studies included in the differential-effect analysis was assessed independently by two reviewers, with discrepancies resolved through discussion. Risk of bias was assessed using Risk of Bias 2 (ROB 2) [26] for randomised trials and Risk of Bias in Non-randomised Studies of Interventions (ROBINS-I) [27] for non-randomised studies. These tools were used to assess bias across their respective domains and to assign an overall risk-of-bias judgment for each study. Data synthesis Owing to the heterogeneity in the studies’ methods, a quantitative synthesis of review results, including meta-analysis, was considered inappropriate. A narrative synthesis was therefore undertaken. First, all included studies that used PROGRESS factors either to describe participants’ baseline characteristics or to assess intervention effects were quantified separately by PROGRESS factor. Second, studies that examined differential intervention effects according to at least one PROGRESS factor were narratively synthesised and presented in tables summarising key study, sample, and intervention characteristics, as well as the methods and findings of the differential-effect analyses. To prepare the tables, included studies were categorised by intervention type and outcome category, and information on measurement tools was extracted. For studies comparing PROGRESS-defined subgroups, overall study outcomes were summarised first, followed by subgroup analyses by PROGRESS factor, with subgroup results and direction of effects reported (Supplementary file, Table 3). For studies conducted exclusively in populations defined by one or more PROGRESS characteristics, outcomes were summarised according to intervention and outcome categories, and primary and secondary outcomes were classified as decreasing, non-significant, or increasing inequity (Supplementary file, Table 4). Before synthesis, interventions, outcomes, measurement tools, and results for primary and secondary outcomes were charted. Results Study selection The search identified 7250 records, which were reduced to 5892 after duplicate removal. Of these, 235 studies were public health or healthcare intervention studies with a comparison design (experimental or quasi-experimental) that reported outcomes. Sixty-six studies reported at least one PROGRESS aspect and were included in this review. Figure 1 presents the study selection process.Fig. 1PRISMA flow diagram of the study selection process PRISMA flow diagram of the study selection process Characteristics of included studies The main characteristics of the 66 included studies are shown in Tables 1 and 2. Although no restriction was placed on publication year, the earliest included study was published in 2001, and approximately two-thirds of studies were published from 2018 onwards. The average sample size was 438 participants (standard deviation, SD: 1195); after excluding two outlier studies with more than 3000 [28] and 9000 [29] participants, respectively, the average sample size was 250 participants (SD 239). Most studies included older adults, with a mean age of 71.4 years (SD 11.1). Target populations varied widely and included people with chronic physical conditions, mental health conditions, disability, multimorbidity, frailty, and healthy older populations.Table 1Overview of included studiesStudy Characteristic (N = 66)N (%)Year (range)2001–2024Sample size:Total sample (average±SD)438 ± 1195Excluding 2 outliers* (average±SD)250 ± 239Age (mean, SD)71.4 (11.1)Design:RCT28 (42.0)Cluster RCT5 (7.5)Quasi-experimental33 (50.0)Level of interventionIndividual18 (27.0)Group/Community46 (70.0)System2 (3.0)Continent:North America35 (53.0)Europe14 (21.0)Asia7 (11.0)Australasia4 (6.0)Africa3 (4.5)South & Central America3 (4.5)Health condition:Chronic conditions (incl. multimorbidity)16 (24)Mental health6 (9)Disability8 (12)Frailty/functional8 (12)Healthy5 (8)Not specified23 (35)* Two studies had sample sizes of over 3000 and over 9000 participants, which were outliers among all other studiesTable 2Descriptive characteristics of included studiesAuthorYearCountryStudy designSample size (intervention/control)Age of participants: mean years ± SD and/or (range)Health condition of participantsLevel of interventionResults NOT reported by PROGRESS factorsPROGRESS factor used only to describe participants (N = 18)Cintoli [30]2021ItalyRCT113 (55/58)74.9 ± 4.7Cognitive disabilityCommunityClifford et al. [31]2024UKCluster RCT100 (56/44)74 ± 5Physically capableCommunityCohen et al. [32]2006USQuasi-experimental166 (90/76)79HealthyCommunityGustafsson et al. [33]2017SwedenRCT416 (227/189)81 (70–97)HealthyCommunityHawkins et al. [34]2024USCohort24175Physical disabilityIndividualKamegaya et al. [35]2014JapanRCT52 (26/26)75 ± 6HealthyCommunityKolt et al. [36]2007New ZealandRCT18675 ± 6Low physical activity levelsIndividualLytras et al. [37]2022GreeceRCT150 (75/75)70 (67–74)History of fallsIndividualNebeker et al. [38]2021USRCT44 (21/23)72 ± 4Low physical activity levelsIndividualPonce-Laguardia [39]2020CubaQuasi-experimental356(60+)Mental healthCommunityRantanen et al. [40]2015FinlandRCT121 (60/61)82 ± 6Physical disabilityIndividualRejeski et al. [41]2009USRCT424 (213/211)(70–89)Low physical activity levelsIndividualRoberts et al. [42]2022USQuasi-experimental2471.6 (66–80)Mental healthCommunityRugbeer et al. [43]2017South AfricaQuasi-experimental8371–72Not specifiedCommunitySmith et al. [44]2018USRCT181 (72/109)48 ± 10 (23–72)MultimorbidityCommunitySun et al. [45]2022ChinaRCT396 (198/198)65.8 ± 12.5 (31–97)RehabilitationIndividualTainta et al. [46]2024SpainRCT125 (61/64)76 ± 6Dementia riskCommunityWillemse et al. [47]2009NetherlandsCohort3667 ± 9 (51–83)Mental healthCommunityPROGRESS factor used as a control variable in multivariable analyses (N = 7)Boen [48]2012NorwayRCT92 (77/61)76Not specifiedCommunityBorson et al. [49]2019USQuasi-experimental21170 ± 11Mental healthIndividualGallucci et al. [50]2020ItalyQuasi-experimental207 (46/161)74.6 ± 5.5Cognitive disabilityCommunityLevine et al. [51]2018USRCT296 (156/140)80.8 ± 8.3Frail older adultsIndividualOry et al. [52]2018USQuasi-experimental430 (163/267)73.3 ± 7.54Not specifiedCommunityPenninx et al. [53]2001USRCT250 (170/80)69.9 ± 5.5OsteoarthritisCommunityTeh et al. [54]2022New ZealandRCT468 (353/115)80 ± 5Pre-frailtyCommunityResults reported by PROGRESS factorsStudy’s outcome reported by PROGERESS factor - Comparison Studies (N = 12)Bay et al. [55]2020USQuasi-experimental12070 ± 8MultimorbidityCommunityCarlson [56]2015USRCT111 (58/53)67.2 ± 6.1Not specifiedCommunityCollins et al. [57]2006UKQuasi-experimental33973.2 ± 8.64 (52–93)Not specifiedCommunityCrist et al. [58]2022USCluster RCT476 (267/209)71 ± 8.9HealthyCommunityHarada [59]2022JapanRCT393 (196/197)74 ± 6.5Not specifiedCommunityKatzman et al. [60]2017USRCT10170 5.7HyperkyphosisIndividualKerr et al. [61]2018USCluster RCT307 (151/156)81.9 ± 5.9Not specifiedCommunityKhunti et al. [62]2021UKRCT1366 (904/460)59 ± 9MetabolicIndividualLang et al. [63]2016FranceQuasi-experimental23266 ± 6.5Mixed health statusCommunityLi et al. [29]2024USQuasi-experimental914774 ± 7Not specifiedSystemLoyola et al. [64]2018ChileQuasi-experimental16474 ± 6Not specifiedCommunitySmith et al. [65]2013USQuasi-experimental13673.3 ± 8.6MultimorbidityCommunityEntire study conducted in a PROGRESS defined population - Population-focused studies (N = 29)Batik et al. [66]2008USRCT1472 ± 7MetabolicIndividualBlancafort Alias et al. [67]2021SpainRCT390 (195/195)73.6 ± 6.9Low self-perceived healthCommunityChapin et al. [68]2013USQuasi-experimental3276 (64–87)Mental healthIndividualChoi et al. [69]2020USRCT8974 ± 9Homebound adultsIndividualCounsell [70]2007USCluster RCT951 (474/477)71.7 ± 5.7Not specifiedCommunityConwell et al. [71]2021USRCT369 (190/179)71 ± 7.5 (60–97)Mental healthCommunityCzaja [72]2018USRCT300 (150/150)76.15 ± 7.4 (65–98)Not specifiedCommunityGeffen et al. [73]2019South AfricaCohort21269Not specifiedIndividualHau et al. [74]2016USCohort5068.4 ± 6.2 (60–85)Physical disabilityCommunityJo et al. [75]2018CanadaCohort7974.1Not specifiedCommunityJoseph et al. [76]2021USQuasi-experimental2056.2 ± 4.3Cardiovascular; MetabolicIndividualKatigbak et al. [77]2024USQuasi-experimental60 (32/28)78.97 ± 4.48 (63–92)HealthyCommunityKhen Siang Ng et al. [78]2020SingaporeCluster RCT378 (164/214)72.19 ± 8.03 (55–95)MultimorbidityCommunityLee et al. [79]2022South KoreaQuasi-experimental124 (62/62)75.2 ± 5.2Not specifiedCommunityLiu et al. [80]2021USRCT42776 ± 7.6 (65–95)Physical disabilityIndividualLukaszyk et al. [81]2017AustraliaCohort9864.0 ± 9.4Not specifiedCommunityMarquez et al. [82]2022USRCT333 (167/166)65 ± 7Low physical activity levelsCommunityMaxwell et al. [83]2022USCohort2068.8 ± 6.6Not specifiedCommunityMendoza-Ruvalcaba & Fernandez-Ballesteros [84]2016MexicoQuasi-experimental76 (50/26)64 ± 4.6Not specifiedCommunityMullins et al. [85]2020USCohort262(69–78)Mental health; MultimorbidityCommunityNgiam et al. [86]2022SingaporeQuasi-experimental138 (84/54)72.3 (70.4–74.2)Not specifiedIndividualNoh et al. [87]2021South KoreaQuasi-experimental877 (331/546)239 participants < 80; 92 participants > 80Mental health; MultimorbidityCommunityOetzel et al. [88]2024New ZealandQuasi-experimental, clustered8672.4 ± 8.2Not specifiedCommunityPolonijo et al. [89]2024USQuasi-experimental2062.7 (51–88)HIVCommunityRosenberg et al. [28]2023South AfricaRCT3,568 (1,785/1,783)54.4 ± 12.3 (40–108)Not specifiedSystemSchmidt et al. [90]2021DenmarkQuasi-experimental32965+Not specifiedCommunitySchwei et al. [91]2021USCohort448 (217/231)80 ± 9Not specifiedCommunityWalsh et al. [92]2024USRCT6158.1 ± 6.3HIVCommunityWellman et al. [93]2007CanadaCohort62074.6 ± 7.5Not specifiedCommunitySD = standard deviation; RCT = randomized controlled trial Table 1Overview of included studiesStudy Characteristic (N = 66)N (%)Year (range)2001–2024Sample size:Total sample (average±SD)438 ± 1195Excluding 2 outliers* (average±SD)250 ± 239Age (mean, SD)71.4 (11.1)Design:RCT28 (42.0)Cluster RCT5 (7.5)Quasi-experimental33 (50.0)Level of interventionIndividual18 (27.0)Group/Community46 (70.0)System2 (3.0)Continent:North America35 (53.0)Europe14 (21.0)Asia7 (11.0)Australasia4 (6.0)Africa3 (4.5)South & Central America3 (4.5)Health condition:Chronic conditions (incl. multimorbidity)16 (24)Mental health6 (9)Disability8 (12)Frailty/functional8 (12)Healthy5 (8)Not specified23 (35)* Two studies had sample sizes of over 3000 and over 9000 participants, which were outliers among all other studies Overview of included studies * Two studies had sample sizes of over 3000 and over 9000 participants, which were outliers among all other studies Table 2Descriptive characteristics of included studiesAuthorYearCountryStudy designSample size (intervention/control)Age of participants: mean years ± SD and/or (range)Health condition of participantsLevel of interventionResults NOT reported by PROGRESS factorsPROGRESS factor used only to describe participants (N = 18)Cintoli [30]2021ItalyRCT113 (55/58)74.9 ± 4.7Cognitive disabilityCommunityClifford et al. [31]2024UKCluster RCT100 (56/44)74 ± 5Physically capableCommunityCohen et al. [32]2006USQuasi-experimental166 (90/76)79HealthyCommunityGustafsson et al. [33]2017SwedenRCT416 (227/189)81 (70–97)HealthyCommunityHawkins et al. [34]2024USCohort24175Physical disabilityIndividualKamegaya et al. [35]2014JapanRCT52 (26/26)75 ± 6HealthyCommunityKolt et al. [36]2007New ZealandRCT18675 ± 6Low physical activity levelsIndividualLytras et al. [37]2022GreeceRCT150 (75/75)70 (67–74)History of fallsIndividualNebeker et al. [38]2021USRCT44 (21/23)72 ± 4Low physical activity levelsIndividualPonce-Laguardia [39]2020CubaQuasi-experimental356(60+)Mental healthCommunityRantanen et al. [40]2015FinlandRCT121 (60/61)82 ± 6Physical disabilityIndividualRejeski et al. [41]2009USRCT424 (213/211)(70–89)Low physical activity levelsIndividualRoberts et al. [42]2022USQuasi-experimental2471.6 (66–80)Mental healthCommunityRugbeer et al. [43]2017South AfricaQuasi-experimental8371–72Not specifiedCommunitySmith et al. [44]2018USRCT181 (72/109)48 ± 10 (23–72)MultimorbidityCommunitySun et al. [45]2022ChinaRCT396 (198/198)65.8 ± 12.5 (31–97)RehabilitationIndividualTainta et al. [46]2024SpainRCT125 (61/64)76 ± 6Dementia riskCommunityWillemse et al. [47]2009NetherlandsCohort3667 ± 9 (51–83)Mental healthCommunityPROGRESS factor used as a control variable in multivariable analyses (N = 7)Boen [48]2012NorwayRCT92 (77/61)76Not specifiedCommunityBorson et al. [49]2019USQuasi-experimental21170 ± 11Mental healthIndividualGallucci et al. [50]2020ItalyQuasi-experimental207 (46/161)74.6 ± 5.5Cognitive disabilityCommunityLevine et al. [51]2018USRCT296 (156/140)80.8 ± 8.3Frail older adultsIndividualOry et al. [52]2018USQuasi-experimental430 (163/267)73.3 ± 7.54Not specifiedCommunityPenninx et al. [53]2001USRCT250 (170/80)69.9 ± 5.5OsteoarthritisCommunityTeh et al. [54]2022New ZealandRCT468 (353/115)80 ± 5Pre-frailtyCommunityResults reported by PROGRESS factorsStudy’s outcome reported by PROGERESS factor - Comparison Studies (N = 12)Bay et al. [55]2020USQuasi-experimental12070 ± 8MultimorbidityCommunityCarlson [56]2015USRCT111 (58/53)67.2 ± 6.1Not specifiedCommunityCollins et al. [57]2006UKQuasi-experimental33973.2 ± 8.64 (52–93)Not specifiedCommunityCrist et al. [58]2022USCluster RCT476 (267/209)71 ± 8.9HealthyCommunityHarada [59]2022JapanRCT393 (196/197)74 ± 6.5Not specifiedCommunityKatzman et al. [60]2017USRCT10170 5.7HyperkyphosisIndividualKerr et al. [61]2018USCluster RCT307 (151/156)81.9 ± 5.9Not specifiedCommunityKhunti et al. [62]2021UKRCT1366 (904/460)59 ± 9MetabolicIndividualLang et al. [63]2016FranceQuasi-experimental23266 ± 6.5Mixed health statusCommunityLi et al. [29]2024USQuasi-experimental914774 ± 7Not specifiedSystemLoyola et al. [64]2018ChileQuasi-experimental16474 ± 6Not specifiedCommunitySmith et al. [65]2013USQuasi-experimental13673.3 ± 8.6MultimorbidityCommunityEntire study conducted in a PROGRESS defined population - Population-focused studies (N = 29)Batik et al. [66]2008USRCT1472 ± 7MetabolicIndividualBlancafort Alias et al. [67]2021SpainRCT390 (195/195)73.6 ± 6.9Low self-perceived healthCommunityChapin et al. [68]2013USQuasi-experimental3276 (64–87)Mental healthIndividualChoi et al. [69]2020USRCT8974 ± 9Homebound adultsIndividualCounsell [70]2007USCluster RCT951 (474/477)71.7 ± 5.7Not specifiedCommunityConwell et al. [71]2021USRCT369 (190/179)71 ± 7.5 (60–97)Mental healthCommunityCzaja [72]2018USRCT300 (150/150)76.15 ± 7.4 (65–98)Not specifiedCommunityGeffen et al. [73]2019South AfricaCohort21269Not specifiedIndividualHau et al. [74]2016USCohort5068.4 ± 6.2 (60–85)Physical disabilityCommunityJo et al. [75]2018CanadaCohort7974.1Not specifiedCommunityJoseph et al. [76]2021USQuasi-experimental2056.2 ± 4.3Cardiovascular; MetabolicIndividualKatigbak et al. [77]2024USQuasi-experimental60 (32/28)78.97 ± 4.48 (63–92)HealthyCommunityKhen Siang Ng et al. [78]2020SingaporeCluster RCT378 (164/214)72.19 ± 8.03 (55–95)MultimorbidityCommunityLee et al. [79]2022South KoreaQuasi-experimental124 (62/62)75.2 ± 5.2Not specifiedCommunityLiu et al. [80]2021USRCT42776 ± 7.6 (65–95)Physical disabilityIndividualLukaszyk et al. [81]2017AustraliaCohort9864.0 ± 9.4Not specifiedCommunityMarquez et al. [82]2022USRCT333 (167/166)65 ± 7Low physical activity levelsCommunityMaxwell et al. [83]2022USCohort2068.8 ± 6.6Not specifiedCommunityMendoza-Ruvalcaba & Fernandez-Ballesteros [84]2016MexicoQuasi-experimental76 (50/26)64 ± 4.6Not specifiedCommunityMullins et al. [85]2020USCohort262(69–78)Mental health; MultimorbidityCommunityNgiam et al. [86]2022SingaporeQuasi-experimental138 (84/54)72.3 (70.4–74.2)Not specifiedIndividualNoh et al. [87]2021South KoreaQuasi-experimental877 (331/546)239 participants < 80; 92 participants > 80Mental health; MultimorbidityCommunityOetzel et al. [88]2024New ZealandQuasi-experimental, clustered8672.4 ± 8.2Not specifiedCommunityPolonijo et al. [89]2024USQuasi-experimental2062.7 (51–88)HIVCommunityRosenberg et al. [28]2023South AfricaRCT3,568 (1,785/1,783)54.4 ± 12.3 (40–108)Not specifiedSystemSchmidt et al. [90]2021DenmarkQuasi-experimental32965+Not specifiedCommunitySchwei et al. [91]2021USCohort448 (217/231)80 ± 9Not specifiedCommunityWalsh et al. [92]2024USRCT6158.1 ± 6.3HIVCommunityWellman et al. [93]2007CanadaCohort62074.6 ± 7.5Not specifiedCommunitySD = standard deviation; RCT = randomized controlled trial Descriptive characteristics of included studies SD = standard deviation; RCT = randomized controlled trial Reporting and analytical use of PROGRESS factors Table 3 summarises how PROGRESS factors were reported and used across included studies. Of the 66 studies, 25 did not analyse outcomes according to PROGRESS factors: 18 (27%) used them only to describe baseline participant characteristics and 7 (11%) included PROGRESS factors only as control variables in multivariable analyses. The remaining 41 (62%) studies analysed the results of the interventions by at least one PROGRESS factor. On average, 4 PROGRESS aspects were reported per study (SD = 1.3). Gender/sex was reported in all studies, followed by education (N = 49, 74%), race/ethnicity (N = 38, 58%), and socioeconomic status (N = 38, 57.5%), whereas religion was the least frequently reported aspect (N = 6, 9%). Socioeconomic status was the PROGRESS aspect most often used in outcome analyses (N = 20, 53%), followed by race/ethnicity (N = 13, 34%). No studies analysed intervention outcomes by occupation or religion.Table 3Distribution and analysis of PROGRESS aspects in included studiesAuthorPROGR.1ESS.1Participant demographics described by PROGRESS factorsCintoli [30]○○Clifford et al. [31]○○○○○○Cohen et al. [32]○○Gustafsson et al. [33]○○○○Hawkins et al. [34]○○○Kamegaya et al. [35]○○Kolt et al. [36]○○○○○Lytras et al. [37]○○Nebeker et al. [38]○○Ponce-Laguardia [39]○○○Rantanen et al. [40]○○Rejeski et al. [41]○○○Roberts et al. [42]○○○Rugbeer et al. [43]○○○Smith et al. [44]○○○Sun et al. [45]○○○○○Tainta et al. [46]○○○Willemse et al. [47]○○○PROGRESS factor used as a control variable in multivariable analysesBoen [48]○◐○○◐Borson et al. [49]◐◐◐Gallucci et al. [50]◐◐Levine et al. [51]◐◐○○○Ory et al. [52]◐◐◐○Penninx et al. [53]◐◐○○Teh et al. [54]◐◐○◐○Results Analysed by PROGRESS FactorsComparison StudiesBay et al. [55]●○○○○○Carlson [56]○●◐Collins et al. [57]○●○●●Crist et al. [58]◐○●○◐●Harada [59]●●●●Katzman et al. [60]●○○Kerr et al. [61]●○◐Khunti et al. [62]●●◐●Lang et al. [63]○○●○○Li et al. [29]○○◐●●◐Loyola et al. [64]○○●Smith et al. [65]●○○○Population-focused StudiesBatik et al. [66]○○●Blancafort Alias et al. [67]●○○●○Chapin et al. [68]○○○○●○Choi et al. [69]○○●●Conwell et al. [71]○○○○●Counsell [70]○●○●○Czaja [72]○○○●●Geffen et al. [73]○○○●○Hau et al. [74]●●Jo et al. [75]●○○Joseph et al. [76]●○○○○Katigbak et al. [77]●○○○○Khen Siang Ng et al. [78]●○○○●●Lee et al. [79]●○○○○Liu et al. [80]◐◐○●○Lukaszyk et al. [81]●○○Marquez et al. [82]●○○○○Maxwell et al. [83]○○○○●○Mendoza-Ruvalcaba & Fernandez-Ballesteros [84]●○○○Mullins et al. [85]○○●Ngiam et al. [86]○◐◐●◐Noh et al. [87]◐○○●Oetzel et al. [88]●○○Polonijo et al. [89]●○●○○●Rosenberg et al. [28]●●●Schmidt et al. [90]○○●○Schwei et al. [91]○◐◐○○●Walsh et al. [92]●○○○○○Wellman et al. [93]○○○○●○Key: Open circle = Reported as descriptive only, no analysis done; Half-shaded circle = Used in the analyses as control variable; Filled circle = Results of the intervention are analyses by this factor; P = place of residence; R = Race/ethnicity/culture; O = Occupation; G = Gender/sex; R.1 = Religion; E = Education; S = Socioeconomic Status; S.1 = Social capital Table 3Distribution and analysis of PROGRESS aspects in included studiesAuthorPROGR.1ESS.1Participant demographics described by PROGRESS factorsCintoli [30]○○Clifford et al. [31]○○○○○○Cohen et al. [32]○○Gustafsson et al. [33]○○○○Hawkins et al. [34]○○○Kamegaya et al. [35]○○Kolt et al. [36]○○○○○Lytras et al. [37]○○Nebeker et al. [38]○○Ponce-Laguardia [39]○○○Rantanen et al. [40]○○Rejeski et al. [41]○○○Roberts et al. [42]○○○Rugbeer et al. [43]○○○Smith et al. [44]○○○Sun et al. [45]○○○○○Tainta et al. [46]○○○Willemse et al. [47]○○○PROGRESS factor used as a control variable in multivariable analysesBoen [48]○◐○○◐Borson et al. [49]◐◐◐Gallucci et al. [50]◐◐Levine et al. [51]◐◐○○○Ory et al. [52]◐◐◐○Penninx et al. [53]◐◐○○Teh et al. [54]◐◐○◐○Results Analysed by PROGRESS FactorsComparison StudiesBay et al. [55]●○○○○○Carlson [56]○●◐Collins et al. [57]○●○●●Crist et al. [58]◐○●○◐●Harada [59]●●●●Katzman et al. [60]●○○Kerr et al. [61]●○◐Khunti et al. [62]●●◐●Lang et al. [63]○○●○○Li et al. [29]○○◐●●◐Loyola et al. [64]○○●Smith et al. [65]●○○○Population-focused StudiesBatik et al. [66]○○●Blancafort Alias et al. [67]●○○●○Chapin et al. [68]○○○○●○Choi et al. [69]○○●●Conwell et al. [71]○○○○●Counsell [70]○●○●○Czaja [72]○○○●●Geffen et al. [73]○○○●○Hau et al. [74]●●Jo et al. [75]●○○Joseph et al. [76]●○○○○Katigbak et al. [77]●○○○○Khen Siang Ng et al. [78]●○○○●●Lee et al. [79]●○○○○Liu et al. [80]◐◐○●○Lukaszyk et al. [81]●○○Marquez et al. [82]●○○○○Maxwell et al. [83]○○○○●○Mendoza-Ruvalcaba & Fernandez-Ballesteros [84]●○○○Mullins et al. [85]○○●Ngiam et al. [86]○◐◐●◐Noh et al. [87]◐○○●Oetzel et al. [88]●○○Polonijo et al. [89]●○●○○●Rosenberg et al. [28]●●●Schmidt et al. [90]○○●○Schwei et al. [91]○◐◐○○●Walsh et al. [92]●○○○○○Wellman et al. [93]○○○○●○Key: Open circle = Reported as descriptive only, no analysis done; Half-shaded circle = Used in the analyses as control variable; Filled circle = Results of the intervention are analyses by this factor; P = place of residence; R = Race/ethnicity/culture; O = Occupation; G = Gender/sex; R.1 = Religion; E = Education; S = Socioeconomic Status; S.1 = Social capital Distribution and analysis of PROGRESS aspects in included studies Key: Open circle = Reported as descriptive only, no analysis done; Half-shaded circle = Used in the analyses as control variable; Filled circle = Results of the intervention are analyses by this factor; P = place of residence; R = Race/ethnicity/culture; O = Occupation; G = Gender/sex; R.1 = Religion; E = Education; S = Socioeconomic Status; S.1 = Social capital Intervention outcomes Among the 41 studies that reported intervention effects by at least one PROGRESS factor, a total of 254 outcomes were identified. Of these 254 outcomes, 98 (38.6%) suggested more favourable effects or reduced inequity in PROGRESS-defined populations, 146 (57.5%) showed no statistically significant differential effect, and 10 (3.9%) suggested less favourable effects or increased inequity. Table 4 summarizes the outcome effect on inequity by intervention type, and Supplementary Table 3 and 4 provides detailed study-level outcome data.Table 4Summary of outcome effect on inequity by intervention typeOutcomeInterventionQuality of lifeFunctioningClinicalHealth Services/SatisfactionSelf EfficacyTotalDecreaseNullIncreaseDecreaseNullIncreaseDecreaseNullIncreaseDecreaseNullIncreaseDecreaseNullIncreaseDecreaseNullIncreaseN (%)N (%)N (%)N (%)N (%)N (%)N (%)N(%)N (%)N (%)N (%)N (%)N (%)N (%)N (%)N (%)N (%)N (%)Physical activity **5 (56)4 (44)04 (44)5 (56)033 (44)38(51)4 (5)00001 (100)042 (45)48 (51)4 (4)Lifestyle behaviours9 (45)11(55)05 (38)8 (62)05 (28)11(61)2 (11)2 (67)1 (33)03 (100)0024 (42)31 (54)2 (4)Social behaviours4 (18)17(77)1 (5)6 (20)21(70)3 (10)00001 (100)04 (57)3 (43)014 (23)42 (70)4 (7)Psychological wellbeing7 (33)14(67)06 (60)4 (40)000000001 (100)013 (41)19 (59)0Medication use/healthcare engagement*1 (33)2 (66)02 (67)1 (33)002(100)01 (50)1 (50)01 (100)005 (45)6 (55)0* Including the study by Rosenberg et al., classified as “healthcare engagement”** most physical activity interventions focus only on physical activity, one included also dietary behavioural aspects Table 4Summary of outcome effect on inequity by intervention typeOutcomeInterventionQuality of lifeFunctioningClinicalHealth Services/SatisfactionSelf EfficacyTotalDecreaseNullIncreaseDecreaseNullIncreaseDecreaseNullIncreaseDecreaseNullIncreaseDecreaseNullIncreaseDecreaseNullIncreaseN (%)N (%)N (%)N (%)N (%)N (%)N (%)N(%)N (%)N (%)N (%)N (%)N (%)N (%)N (%)N (%)N (%)N (%)Physical activity **5 (56)4 (44)04 (44)5 (56)033 (44)38(51)4 (5)00001 (100)042 (45)48 (51)4 (4)Lifestyle behaviours9 (45)11(55)05 (38)8 (62)05 (28)11(61)2 (11)2 (67)1 (33)03 (100)0024 (42)31 (54)2 (4)Social behaviours4 (18)17(77)1 (5)6 (20)21(70)3 (10)00001 (100)04 (57)3 (43)014 (23)42 (70)4 (7)Psychological wellbeing7 (33)14(67)06 (60)4 (40)000000001 (100)013 (41)19 (59)0Medication use/healthcare engagement*1 (33)2 (66)02 (67)1 (33)002(100)01 (50)1 (50)01 (100)005 (45)6 (55)0* Including the study by Rosenberg et al., classified as “healthcare engagement”** most physical activity interventions focus only on physical activity, one included also dietary behavioural aspects Summary of outcome effect on inequity by intervention type * Including the study by Rosenberg et al., classified as “healthcare engagement” ** most physical activity interventions focus only on physical activity, one included also dietary behavioural aspects Findings by intervention type Physical activity There were 16 studies with physical activity interventions, of which one was a mixed intervention combining physical activity with diet and nutrition [93]. These studies examined the greatest number of outcomes, mainly clinical outcomes (N = 75), followed by quality of life (N = 9), and functioning (N = 9). Of all outcomes examined, 42 (45%) were favourable for the PROGRESS population, indicating decreased inequity; 48 (51%) were non-significant, and 4 (4%) showed unfavourable findings, indicating increased inequity. Among clinical outcomes, less than half (N = 33, 44%) showed decreased inequities. Examples for decreased inequity include a study showing greater improvement in steps per day among participants living in more deprived areas [62], a study indicating greater improvement in disability and mobility outcomes among groups with lower social capital [64], and a study of a health education and research participation enhancement program that showed improved social functioning among Black participants [55], while 38 (51%) were non-significant. Four clinical outcomes (5%) showed unfavourable effects, indicating increased inequity, including one study suggesting greater benefit among men, college-educated participants, and married participants [61], and another study in a low socioeconomic status population reporting worse sleep quality [83]. Nonetheless, the latter study reported mostly non-significant findings for other clinical outcomes, including BMI, fatigue, and perceived exertion, alongside some positive findings such as improved strength and flexibility. Lifestyle behaviour A total of 9 studies examined lifestyle behaviour interventions. These studies examined mostly quality of life (N = 20), clinical (N = 18), and functioning outcomes (N = 13). Of all outcomes examined, 24 (42%) showed favourable findings for the PROGRESS group, indicating decreased inequity, 31 (54%) were non-significant, and 2 (4%) showed increased inequity. Of interventions examining quality of life outcomes, less than half (N = 9, 45%) showed decreased inequities, while 11 (55%) were non-significant. Positive quality-of-life findings included greater reductions in unhealthy physical and general unhealthy days among Hispanic and African American participants than among White participants in one study [44], and improved physical, general, and mental quality of life in a low-socioeconomic-status population in another [70]. Among clinical outcomes, 5 (28%) were positive for the PROGRESS group, indicating decreased inequity, 11 (61%) were non-significant, and 2 (11%), showed increased inequity. Both unfavourable outcomes came from one study examining cortical and hippocampal volume, where the intervention was associated with significant increases in men but not in women, suggesting increased inequity by sex [56]. Positive clinical findings were also reported in one study of a low-socioeconomic-status population, which found improved hospital admissions among participants at high risk of readmission [70]. Similarly, studies examining functioning outcomes showed mostly non-significant results (N = 8, 62%), with 5 (38%) indicating decreased inequity. Positive functioning findings included improved frailty in older adults living alone [87] and improved satisfaction with social relationships in another study [84]. Social connectedness There were 9 studies examining social connectedness interventions. These studies mainly examined outcomes of functioning (N = 30) and quality of life (N = 22). Of all outcomes examined, 14 (23%) indicated decreased inequity for the PROGRESS population, 42 (70%) had non-significant effects, and 4 (7%) indicated worse outcomes, with increased inequity among the PROGRESS-defined population. Of the functioning outcomes, 21 (70%) were non-significant, 6 (20%) showed favourable effects, indicating decreased inequity, and 3 (10%) showed unfavourable effects. Unfavourable functioning outcomes included one study suggesting increased inequity in loneliness by education and socioeconomic status [29], one study reporting decreased resilience in the PROGRESS-defined population [91], and one study showing mixed effects, with reduced inequity by income but increased inequity by education within minority participants [57]. Positive functioning outcomes were also reported, including improved social interaction, satisfaction with social support, and reduced social isolation in homebound, socially isolated older adults [69], and improved social engagement in another study [89]. Of the quality-of-life outcomes examined, 4 (18%) showed favourable effects, 17 (77%) were non-significant, and 1 (5%) showed an unfavourable effect, with one study of a peer-to-peer support intervention reporting increased anxiety symptoms for the PROGRESS-defined population [91]. Positive quality of life outcomes included improved mental quality of life following a tele-delivered behavioural activation intervention in homebound, socially isolated older adults [69], improved energy and wellbeing in a technology-based intervention designed to enhance social support and reduce isolation [94], and improved negative affect in a brief intervention to increase social engagement among socially isolated older adults with HIV [89]. Psychological wellbeing A total of 5 studies examined psychological wellbeing interventions, mainly by providing mental health peer support and educational programmes aimed at improving wellbeing. These studies mainly examined quality-of-life outcomes (N = 21, 66%) and functioning outcomes (N = 10, 31%), with one additional self-efficacy outcome. Overall, 13 outcomes (41%) showed favourable findings for the PROGRESS group, indicating decreased inequity, while 19 outcomes (59%) were non-significant; no unfavourable findings were reported. Quality-of-life outcomes were favourable in 7 (33%) cases, and non-significant in 14 (67%). Positive quality-of-life findings included improvements in depression and one quality-of-life subscale [68], improved overall wellbeing [73], improved mental quality of life [71], and improvements in selected health-related quality-of-life subscales [75]. Functioning outcomes were favourable in 6 (60%) cases and were non-significant in 4 (40%). Positive functioning findings included improved perceived social support [73] and reduced burden [71]. No negative outcomes were reported in studies of interventions aimed at improving psychological wellbeing. Medication use/healthcare engagement Interventions targeting medication use or healthcare engagement were the least examined, with only one study [92]. This study pilot-tested four remotely delivered interventions for rural older adults living with HIV in the southern United States: group-based social support, group-based stigma reduction, individual strengths-based case management, and individual technology detailing. Outcomes examined included quality of life (N = 3), functioning (N = 3), clinical outcomes (N = 2), health services/satisfaction (N = 1), and self-efficacy (N = 1). Overall, 5 outcomes (45%) showed favourable findings for the PROGRESS group and 6 (55%) were non-significant, with no unfavourable findings reported. Positive findings included higher odds of perfect medication adherence among participants assigned to the social support intervention compared with those not assigned to it, improved personal-beliefs quality-of-life scores among participants assigned to strengths-based case management compared with those not assigned, reduced loneliness and internalised stigma among participants assigned to the stigma-reduction or social-support interventions, and improved eHealth literacy among participants assigned to the technology-detailing intervention [92]. Other specified target One study [28] examined an intervention classified as other. This study evaluated a cash-transfer intervention in a rural population and found an overall non-significant effect on the primary clinical outcome. However, when intersectionality was examined, worse outcomes were reported among participants with lower education and lower income, suggesting increased inequity within the PROGRESS-defined group. Methodological quality Forty-one studies examined differential intervention effects by at least one PROGRESS aspect and were therefore considered to include an equity assessment; these comprised 17 randomized studies and 24 non-randomized studies. Across these 41 studies overall methodological quality was moderate to high with only two studies being of low quality. (Supplementary File 5 and 6). Discussion Our systematic review reveals a striking evidence gap: although all included studies reported at least one PROGRESS characteristic, only 41 of 66 analysed intervention outcomes by at least one inequity dimension, and detailed equity analyses remained limited. The scarcity of evidence on how interventions affect socially disadvantaged subgroups is particularly concerning given the well-documented challenges of healthy ageing and the disproportionate burden of multimorbidity in these populations [95–98]. This lack of detailed reporting limits our ability to understand for whom and under what conditions interventions are effective, thereby hindering the development and broad dissemination of equity-oriented healthy ageing strategies. The limited evidence available on the effects of healthy ageing interventions across PROGRESS dimensions presents a mixed picture. Across all extracted outcomes, most findings were non-significant, but some suggested decreased inequity and a smaller number suggested increased inequity. Gender/sex was reported in all included studies, yet only two [89, 92] explicitly distinguished sex from gender, underscoring the scarcity of data that meaningfully accounts for gender diversity and the need for more inclusive analyses of intervention effects across diverse gender identities, including LGBTQ+ populations. Intersectional analyses were also rare, despite the likelihood that multiple, overlapping forms of disadvantage shape intervention uptake and effectiveness. These patterns may also partly reflect publication bias, favouring the publication of studies with non-negative findings [99], and thereby underrepresenting interventions that worsen inequities or show no benefit in disadvantaged populations. This evidence presents the troubling risk that, for specific groups (e.g., racial or ethnic minorities) as compared with the overall population, inequity widens rather than diminishes when healthy ageing interventions are implemented. For example, Kerr et al. [61] demonstrated that although a physical activity intervention improved physical performance, the subgroup findings suggested greater benefit among men, college-educated participants, and married participants, indicating possible increased inequity. Likewise, Li et al. [29] found non-significant overall effects of a social connectedness intervention, yet reductions in loneliness and increases in social activities were observed only among older adults with higher education and income, thereby exacerbating inequities. The authors emphasise that uptake may differ across groups depending on the type of intervention, an important consideration for large-scale implementation. Social connectedness interventions may be particularly susceptible to widening inequities, as their effectiveness often depends on uptake, sustained engagement, and the ability to mobilise existing social or material resources, which may be more available to participants with higher educational or socioeconomic advantage. Taken together, these examples add to the literature on how unintended increases in inequity are observed across intervention types [100, 101], underscoring the need for context-specific measures to ensure interventions are generalisable and equitable across diverse populations [55, 102]. At the same time, the review also suggests that healthy ageing interventions can support more equitable outcomes when they are responsive to the needs of disadvantaged groups. Several studies reported greater gains among participants living in more deprived areas, with lower social capital, or from racial and ethnic minority groups, indicating that inequities are not inevitably widened and may be reduced under some conditions. However, interpreting these patterns requires caution. Many studies [55, 56, 61] did not report baseline and follow-up outcome scores in sufficient detail and often relied on within-group rather than between-group analyses. Because disadvantaged populations frequently begin with poorer baseline outcomes [97, 103], greater improvement alone does not necessarily indicate that inequity has been reduced. More rigorous use of between-group and difference-in-differences analyses is needed to determine whether observed gains reflect true narrowing of social gaps. This need has also been highlighted in a recent Cochrane review examining how health equity effects are assessed in systematic reviews of interventions [104]. These findings have important implications for future intervention planning. Explicitly integrating equity dimensions, particularly underrepresented factors such as occupation, religion, gender identities and place of residence, can improve both the inclusivity and effectiveness of public health and healthcare interventions. Most included studies evaluated group- or community-level approaches [57–59, 61, 64, 65], whereas individual-level interventions [60, 62, 66, 68, 73, 76, 86] were less common and system-level interventions [28, 29] were rare. This imbalance limits the broader applicability of findings and makes it difficult to generate robust system-level policy recommendations. Moreover, most studies had small sample sizes, limiting the ability to test the effectiveness of interventions in subgroups affected by compound disadvantaged conditions. Future studies should be designed and powered to assess subgroup and intersectional effects, and should embed an equity lens, as recommended by the WHO [105], throughout intervention planning, implementation, and evaluation. This review has limitations that underscore the need for further research. First, considerable heterogeneity among the included studies, in terms of intervention types, study populations and settings, and outcome measures, posed challenges for drawing uniform conclusions. We sought to mitigate this by applying consistent classifications (e.g., the International Classification of Health Interventions (ICHI) for categorizing intervention types). Importantly, because the primary aim of this review was to assess the overall direction of equity effects (i.e., increase, decrease, or no effect) rather than to quantify precise effect sizes, such variability was acceptable. Second, the generally small sample sizes and the modest methodological quality of many included studies limit the strength of the conclusions. Nevertheless, the observation of unintended negative effects in PROGRESS populations underscores the danger of healthy ageing interventions to exacerbate inequities, a possibility that must be carefully considered in the planning and implementation of future interventions. Finally, we restricted our equity assessment to the PROGRESS factors and excluded the additional “Plus” elements of the PROGRESS-Plus framework. This decision, while enhancing conceptual consistency by focusing on social rather than health-related determinants (e.g., age and disability), may have limited our ability to capture certain individual-level vulnerabilities [106]. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material Supplementary Material